The present invention relates to magnetic resonance imaging (MRI) systems, and particularly to the radio frequency (RF) coils used in such systems.
Magnetic resonance imaging (MRI) utilizes hydrogen nuclear spins of the water molecules in the human body or other tissue, which are polarized by a strong, uniform, static magnetic field generated by a magnet (referred to as B0—the main magnetic field in MRI physics). The magnetically polarized nuclear spins generate magnetic moments in the human body. The magnetic moments point in the direction of the main magnetic field in a steady state, and produce no useful information if they are not disturbed by any excitation.
The generation of nuclear magnetic resonance (NMR) signal for MRI data acquisition is achieved by exciting the magnetic moments with a uniform radio frequency (RF) magnetic field (referred to as the B1 field or the excitation field). The B1 field is produced in the imaging region of interest by an RF transmit coil which is driven by a computer-controlled RF transmitter with a power amplifier. During the excitation, the nuclear spin system absorbs magnetic energy, and its magnetic moments precess around the direction of the main magnetic field. After the excitation, the precessing magnetic moments will go through a process of free induction decay, emitting their absorbed energy and then returning to the steady state. During the free induction decay, NMR signals are detected by the use of a receive RF coil, which is placed in the vicinity of the excited volume of the human body. The NMR signal is an induced electrical motive force (voltage), or current, in the receive RF coil that has been induced by the flux change over some time period due to the relaxation of precessing magnetic moments in the human tissue. This signal provides the contrast information of the image. The receive RF coil can be either the transmit coil itself, or an independent receive-only RF coil. The NMR signal is used for producing magnetic resonance images by using additional pulsed magnetic gradient fields, which are generated by gradient coils integrated inside the main magnet system. The gradient fields are used to spatially encode the signals and selectively excite a specific volume of the human body. There are usually three sets of gradient coils in a standard MRI system, which generate magnetic fields in the same direction of the main magnetic field, varying linearly in the imaging volume.
In MRI, it is desirable for the excitation and reception to be spatially uniform in the imaging volume for better image uniformity. In a standard MRI system, the best excitation field homogeneity is usually obtained by using a whole-body volume RF coil for transmission. The whole-body transmit coil is the largest RF coil in the system. A large coil, however, produces lower signal-to-noise ratio (S/N) if it is also used for reception, mainly because of its greater distance from the signal-generating tissues being imaged. Since a high signal-to-noise ratio is the most desirable factor in MRI, special-purpose coils are used for reception to enhance the S/N ratio from the volume of interest.
In practice, a well-designed specialty RF coil should have the following functional properties: high S/N ratio, good uniformity, high unloaded quality factor (Q) of the resonance circuit, and high ratio of the unloaded to loaded Q factors. In addition, the coil device must be mechanically designed to facilitate patient handling and comfort, and to provide a protective barrier between the patient and the RF electronics. Another way to increase the S/N is by quadrature reception. In this method, NMR signals are detected in two orthogonal directions, which are in the transverse plane or perpendicular to the main magnetic field. The two signals are detected by two independent individual coils which cover the same volume of interest. With quadrature reception, the S/N can be increased by up to ✓2 over that of the individual linear coils.
To cover a large field-of-view, while maintaining the S/N characteristic of a small and conformal coil, a linear surface coil array technique was created to image the entire human spines (U.S. Pat. No. 4,825,162). Subsequently, other linear surface array coils were used for C.L. spine imaging, such as the technique described in U.S. Pat. No. 5,198,768. These two devices consist of an array of planar linear surface coil elements. These coil systems do not work well for imaging deep tissues, such as the blood vessels in the lower abdomen, due to sensitivity drop-off away from the coil surface.
To image the lower extremities, quadrature phased array coils have been utilized such as described in U.S. Pat. Nos. 5,430,378 and 5,548,218. The first quadrature phased array coil, images the lower extremities by using two orthogonal linear coil arrays: six planar loop coil elements placed in the horizontal plane and underneath the patient and six planar loop coil elements placed in the vertical plane and in between the legs. Each linear coil array functions in a similar way as described in U.S. Pat. No. 4,825,162 (Roemer). The second quadrature phased array coil (Lu) was designed to image the blood vessels from the pelvis down. This device also consists of two orthogonal linear coil arrays extending in the head-to-toe direction: a planar array of loop coil elements laterally centrally located on top of the second array of butterfly coil elements. The loop coils are placed immediately underneath the patient and the butterfly coils are wrapped around the patient. Again, each linear coil array functions in a similar way as described in U.S. Pat. No. 4,825,162.
In MRI, gradient coils are routinely used to give phase-encoding information to a sample to be imaged. To obtain an image, it is required that all the data points in a so-called “k-space” (i.e., frequency space) must be collected. Recently, there have been developments where some of the data points in k-space are intentionally skipped and at the same time use the intrinsic sensitivity information of RF receive coils as the phase-encoding information for those skipped data points. This action takes place simultaneously, and thus is referred to as partially parallel imaging or partially parallel acquisition (PPA). By collecting multiple data points simultaneously, it requires less time to acquire the same amount of data, when compared with the conventional gradient-only phase-encoding approach. The time savings can be used to reduce total imaging time, in particular, for the applications in which cardiac or respiratory motions in tissues being imaged become concerns, or to collect more data to achieve better resolution or S/N. SiMultaneous Acquisition of Spatial Harmonics, SMASH, (U.S. Pat. No. 5,910,728 and “Simultaneous Acquisition of Spatial Harmonics (SMASH): Fast Imaging with Radiofrequency Coil Arrays,” Daniel K. Sodickson and Warren J. Manning, Magnetic Resonance in Medicine 38:591–603 (1997), both incorporated herein by reference) and “SENSE: Sensitivity Encoding for Fast MRI,” Klaas P. Pruessmann, et al., Magnetic Resonance in Medicine 42:952–962 (1999, also incorporated by reference, are basically two methods of PPA. SMASH takes advantage of the parallel imaging by skipping phase encode lines that yield decreasing the Field-of-View (FOV) in the phase-encoding direction and uses coils (e.g., coil arrays) together with reconstruction techniques to fill in the missing data points in k-space. SENSE, on the other hand, is a technique that utilizes a reduced FOV in the read direction, resulting an aliased image that is then unfolded in x-space (i.e., real space), while using the RF coil sensitivity information, to obtain a true corresponding image. Here, we make use of phase difference between signals from multiple coils to skip phase encoding steps. By skipping some of the phase encoding steps, one can achieve speeding up imaging process by a reduction factor R. Theoretically speaking, the factor R should equal the number of independent coils/arrays. In the SENSE approach, the SIR is defined as:SNRSENSE=SNRFULL/{g✓R}where SNRFULL is the S/N achievable when all the phase encoding steps are collected by traditional gradient phase encoding scheme. SNRSENSE is optimized when the geometry factor g equals 1. To obtain g of 1, traditional decoupling techniques such as overlapping nearest neighbor elements to null the mutual inductance between them shall not apply, as have been reported by others.
SENSE and SMASH or a hybrid approach of both demand a new type of design requirements in RF coil design. In SMASH, the primary criterion for the array is that it be capable of generating sinusoids whose wavelengths are on the order of the FOV. This is how the target FOV along the phase encoding direction for the array is determined. Conventional array designs can incorporate element and array dimensions that will give optimal S/N for the object of interest. In addition, users of conventional arrays are free to choose practically any FOV, as long as severe aliasing artifacts are not a problem. In contrast, when using SMASH, the size of the array determines the approximate range of FOVs that can be used in the imaging experiment. This then determines the approximate element dimensions, assuming complete coverage of the FOV is desired, as in most cases. In SENSE, the method is based upon the fact that the sensitivity of a RF receiver coil generally has a phase-encoding effect complementary to those achieved by linear field gradients. For SENSE imaging, the elements of a coil array should be smaller than for common phased-array imaging, resulting in a trade-off between basic noise and geometry factor, and adjacent coil elements should not overlap for a net gain in S/N due to the improved geometry factor when using SENSE.
For PPA applications, different types of RF coils or arrays have been used so far. However, most of them are based upon “traditional” RF coil design requirements, thus remain within the conventional coil design scheme. It has been reported, however, that since the phase information of B1 of a receive coil is very important when SENSE applications are demanded, for example, new coil design techniques such as non-overlapping adjacent coil elements may be necessary for better definition of the individual phase information associated with each RF coil used in an array, unlike traditional design scheme where two adjacent coils elements are overlapped to null the mutual inductance between the elements (U.S. Pat. No. 4,825,162). Without overlap, the coupling may be increased, but there is a net gain in S/N due to the improved geometry factor when using SENSE. As stated in the above, the use of smaller coil-elements than those for conventional imaging results in a trade-off between basic noise and geometry factor.